In order to study the effectiveness of GPS (Global Positioning System) differential positioning for the deployed Antarctic penetrator, we made fall tests by changing the release altitude at 160, 330, 680, and 1, 000 m above the ground. A two-blade helicopter equipped with a Trimble GPS Pathfinder was used in the experiment at the Aoyama Pasture, Hokkaido, in April 1991. As compared with the precise location (0.1 m accuracy) determined by the GPS doubly differenced phase analysis, post-processed 1 min (less than 60 data points during PDOP < 10) average of the helicopter hovering GPS differential navigation data was accurate to 10 m for a horizontal location (hovering method). By knowing the release time to an accuracy of 1 s, the impact location can be predicted by tracing the falling trajectory (trajectory method). Thus estimated position was accurate to 30 m against the precise location. As for a height accuracy, there was an error of ± 10 m in the hovering method. This error further degraded to ± 20 m when the coordinates of the reference site were replaced by a time-average (1-2 h duration) of the point-positioning results. The above obtained positioning accuracies are enough for a long-range (300 km profile) seismic explosion experiments of each 5-10 km station separation, because the associated errors result only in 0.2% uncertainty on the estimate of P-wave velocity structures.